High spatial resolution thermal conductivity and Raman spectroscopy investigation of hydride vapor phase epitaxy grown n-GaN/sapphire (0001): Doping dependence

Abstract

We have measured high resolution thermal conductivity (K) and Raman spectra {E2 mode [high frequency], A1 mode [longitudinal optical (LO)], and high frequency LO-plasmon coupled mode [LPP+]} at 300 K of three series of n-GaN/sapphire (0001) samples fabricated by hydride vapor phase epitaxy (HVPE). The former was determined with a scanning thermal microscope while the latter was obtained using a micro-Raman system, both having a spatial resolution of ≈2-3 μm. For all three sets of samples the thermal conductivity decreased linearly with log n, about a factor of two decrease in K for every decade increase in n. Also, we found a correlation between film thickness and improved thermal conductivity. Furthermore, K≈ 1.95 W/cm K for one of the most lightly doped samples (≈ 6.9 × 1016 cm-3), higher than previously reported K ≈ 1.7-1.8 W/cm K on lateral epitaxial overgrown (LEO) material with n ≈ (1-2) × 1017 cm-3 [V. M. Asnin et al., Appl. Phys. Lett. 75, 1240 (1999)], K = 1.55 W/cm K on LEO samples using a third-harmonic technique [C. Y. Luo et al., Appl. Phys. Lett. 75, 4151 (1999)], and K ≈ 1.3 W/cm K on a HVPE sample [E. K. Sichel and J. I. Pankove, J. Phys. Chem. Solids 38, 330 (1977)]. The carrier concentration dependence of K is similar to that of other semiconductors in a comparable temperature range. On a log-log scale the linewidth of the observed E2 Raman mode remained constant up to n ≈ 1 × 1018 cm-3 and then increased linearly. The carrier concentration obtained from the LPP+ mode is less than the Hall effect determination. This is probably due to the fact that the latter measures n in both the epilayer and GaN/sapphire interfacial region [D. C. Look and R. J. Molnar, Appl. Phys. Lett. 70, 3377 (1997); W. Götz et al., Appl. Phys. Lett. 72, 1214 (1998)] while the Raman signal is primarily from the epilayer. © 2000 American Institute of Physics.